Analysis of Large Capacity Reversible Data Hiding for ECG Using PEE and Regression

Data security in the field of medical data has raised concerns, particularly when it comes to vital information, such as electrocardiogram (ECG) data. ECG data can provide insights into cardiovascular-related diseases, making it essential to implement special schemes, especially during transmission. In this paper, the use of Prediction Error Expansion with looping is proposed to enhance the capacity for secret data storage, achieving a capacity of up to 0.99 bits per sample for ECG data hiding. This method also maintains reversibility by generating the original ECG signal during the extraction process. To expedite the prediction process, regression algorithms are tested to obtain predictions that closely approximate the original values while ensuring faster computations. Evaluation is carried out by computing the Percentage Residual Difference, Normalized Cross-Correlation, and Signal-to-Noise Ratio. The experimental results show that SVR excels in maintaining signal fidelity but is significantly slower compared to other models. At the same time, ElasticNet and LASSO are 20 times faster but come at the cost of a slightly more pronounced signal discrepancy.